Application of Neural Network to 3D Temperature Measurement by Thermochromic Liquid Crystal

Abstract

Field temperature measurement technique using the flow visualization with the thermochronic liquid crystal has been investigated and is applied to the three-dimensional temperature measurement by scanning the thermometry. The color to temperature calibration of the liquid crystal is carried out by the neural network, where the effect of view angle relative to the light sheet is considered by distributing multi networks with an optimum smoothing technique over the image plane. The calibration error of temperature due to the view angle is much reduced by the present technique. And this technique is proved to be very useful by application to the temperature measurement of thermal convection over a heated surface. The three-dimensional temperature distribution of thermal convention is reconstructed from the measured temperature field and the structure of thermal plumes is examined.